Alex Getmanearly AI signals
Alex Getman @alexgetmancom · · Updated

Tools & updates

Anthropic loosened Fable 5 safeguards

⚡️ Anthropic loosened Fable 5 safeguards

The company updated its safety classifiers and cut false positives by around 85%. Fable 5 will now fall back to Opus 5 far less often for questions about health, symptoms, lab results and biology. Virology, toxicology and molecular design still remain restricted.

Diagram showing Fable's safety classifier boundary shift: initially (A) stricter safeguards blocked benign queries like textbook biology, now (B) the boundary moved right, allowing more legitimate requests while keeping dual-use and harmful content restricted.
Fable 5AnthropicClaude
Русская версияMarkdown

Analysis

Related posts